This is 100% fine as long as it's a good answer. Even if the user only posted the question as a place to put the answer they want to share with everyone else. e.g. as a good canonical duplicate for a frequently-asked question, or just something interesting they wondered and researched themselves.
I've done this a few times myself, on Stack Overflow, to share original research / thoughts / something I figured out (e.g. this, this, or this), or as an FAQ (this) which has proved very useful as a link to leave in a comment, or as a duplicate for closing questions with that bug. All of these have gotten plenty of upvotes and negligible downvotes. A couple times I posted a question and then ended up answering it after a bit more digging, but some of those weren't definite enough to be worth accepting.
A few of those "original research" questions got upvoted comments like
I like these kinds of questions. Keep asking them!
This is [x86] investigative journalism at its best. Thanks!
which is clear evidence of community support for using a self-answered Q&A as kind of a blog post / article to explain or dig into something. Phrasing the question to work as a real question that sets up your answer can be tricky in some cases. :P
I'd expect the Space.SE community to agree with SO on this.
Where this can go wrong is when someone thinks they know the answer and self-accept an answer with misconceptions or based on wrong facts.
The meta answer you're looking a was written back in 2009, when Stack Exchange revolved more strongly around Stack Overflow, the programming site. Questions there are more often not out of simple curiosity, instead motivated by wanting working code. That can change the dynamic somewhat.
The cases of bad self-answers being accepted on SO are usually like this:
- Beginner asks how to do X, including a lot of misconceptions
- Someone else posts an answer that explains some concepts and explains how to solve a problem without a concrete example.
- The question asker writes their own answer with just working code (especially for a whole program) and no explanation of concepts or what was wrong. Often this is still nasty code that just barely works, or worse only happens to work by accident. They accept this answer because it's what they used.
But fortunately Stack Exchange sites only pin the accepted answer to the top when it's written by someone else. We don't need to downvote poor self-answers into oblivion to get them to not show up on top, so usually they just get ignored if they're not good.
Of course actually bad answers should get downvotes, but no more so than if they weren't accepted. In practice I think this tends to be how people really do vote.